ADEPT: Adaptive Dynamic Early-Exit Process for Transformers

2026-01-07 · Sangmin Yoo, Srikanth Malla, Chiho Choi, Wei D. Lu, Joon Hee Choi arxiv

The inference of large language models imposes significant computational workloads, often requiring the processing of billions of parameters. Although early-exit strategies have proven effective in reducing computational demands by halting inference earlier, they apply either to only the first token in the generation phase or at the prompt level in the prefill phase. Thus, the Key-Value (KV) cache for skipped layers remains a bottleneck for subsequent token generation, limiting the benefits of early exit. We introduce ADEPT (Adaptive Dynamic Early-exit Process for Transformers), a novel approach designed to overcome this issue and enable dynamic early exit in both the prefill and generation phases. The proposed adaptive token-level early-exit mechanism adjusts computation dynamically based on token complexity, optimizing efficiency without compromising performance. ADEPT further enhances KV generation procedure by decoupling sequential dependencies in skipped layers, making token-level early exit more practical. Experimental results demonstrate that ADEPT improves efficiency by up to 25% in language generation tasks and achieves a 4x speed-up in downstream classification tasks, with up to a 45% improvement in performance.

📄 PDF Abstract BibTeX arXiv:2601.03700

Code (0)

등록된 구현이 없습니다.

DAISY: Data Adaptive Self-Supervised Early Exit for Speech Representation Models

2024-06-08 · Tzu-Quan Lin, Hung-Yi Lee, Hao Tang

Self-supervised speech models have shown to be useful for various tasks, but their large size limits the use in devices with low computing power and memory. In this work, we explore early exit, an approach for reducing l…

ADEPT: Continual Pretraining via Adaptive Expansion and Dynamic Decoupled Tuning

2025-10-11 · Jinyang Zhang, Yue Fang, Hongxin Ding, Weibin Liao 외 arxiv

Conventional continual pretraining (CPT) for large language model (LLM) domain adaptation often suffers from catastrophic forgetting and limited domain capacity. Existing strategies adopt layer expansion, introducing add…

Continual Pretraining General Knowledge Domain Adaptation

Adaptive Data Exploitation in Deep Reinforcement Learning

2025-01-22 · Mingqi Yuan, Bo Li, Xin Jin, Wenjun Zeng

We introduce ADEPT: Adaptive Data ExPloiTation, a simple yet powerful framework to enhance the **data efficiency** and **generalization** in deep reinforcement learning (RL). Specifically, ADEPT adaptively manages the us…

Computational Efficiency Deep Reinforcement Learning reinforcement-learning Reinforcement Learning+1

VistaDepth: Frequency Modulation With Bias Reweighting For Enhanced Long-Range Depth Estimation

2025-04-21 · Mingxia Zhan, Li Zhang, Xiaomeng Chu, Beibei Wang

Monocular depth estimation (MDE) aims to predict per-pixel depth values from a single RGB image. Recent advancements have positioned diffusion models as effective MDE tools by framing the challenge as a conditional image…

Conditional Image Generation Depth Estimation Image Generation Monocular Depth Estimation

ADEPT: RL-Aligned Agentic Decoding of Emotion via Evidence Probing Tools -- From Consensus Learning to Ambiguity-Driven Emotion Reasoning

2026-02-13 · Esther Sun, Bo-Hao Su, Abinay Reddy Naini, Shinji Watanabe 외 arxiv

Speech Large Language Models (SLLMs) enable high-level emotion reasoning but often produce ungrounded, text-biased judgments without verifiable acoustic evidence. In contrast, self-supervised speech encoders such as WavL…

Emotion Recognition